AMD Phenom II 920 & 940 Full Review

[YukonTrooper]

Not sure why anyone can pick on this guy for spending HIS time to throw out some valid numbers for you guys. Pretty insane that people can be such jerks to someone who didn't have to do this..

Yes, it really is too bad; however, I hope the reviewers realize that it's probably just some seriously frustrated fans venting their emotions about the performance of the chip, at least in relation to the numbers in this review. Nothing more, nothing less. That's the way I see it anyways.

ok.

9950BE will be added only at clock to clock comparison(pii 940 vs 9950 @ 3.0 and deneb with others @ 3.7)

How about adding GTAIV and UT3 ?

Any way to do clock-per-clock comparisons of Q6600, Q9950, and PII 940 @ 3.0GHz? If not, just Q6600 and PII 940 @ 3.0GHz?

[OverClocker_gr]

agena vs deneb @ 3.0

deneb vs kentsfield vs yorkfield vs bloomfield @ 3.7

[YukonTrooper]

GREAT! Glad to see all those chips will do 3.7GHz. Can't wait to see the results. Thanks for your hard work!

[Shintai]

More popcorn and cola. This is priceless.

I just love a good kübler-ross type of show.

[_Lone_Wolf_]

Not sure why anyone can pick on this guy for spending HIS time to throw out some valid numbers for you guys. Pretty insane that people can be such jerks to someone who didn't have to do this..

Unless I've misunderstood OverClocker_gr's role in the linked site it's a professional review on a website bringing in add revenue and so I would think open to constructive criticism.

[Shintai]

Unless I've misunderstood OverClocker_gr's role in the linked site it's a professional review on a website bringing in add revenue and so I would think open to constructive criticism.

Constructive criticism and frustration is 2 different things.

Alot of people seems to have a phailed dream here. Or yet another case of the hypes.

[RPGWiZaRD]

ok.

9950BE will be added only at clock to clock comparison(pii 940 vs 9950 @ 3.0 and deneb with others @ 3.7)

How about adding GTAIV and UT3 ?

Thanks for taking your time to test PII this thoroughly. UT3 and GTAIV would be perfect additions as both are very CPU intensive and both show gains from quad cores.

At least to me these results are exactly what I had expected from PII, nothing suprising here (Except the overclockability then of course).

[Miss Banana]

It has been obvious from the start what strong and weak points the Phenom II will have, this review shows all the weak ones, and none of the strong ones.

It's still a good review, and not biased, but a performance/watt and overclock part added to it will improve things a lot.

I do not really understand how some can act as if this cpu makes amd the biggest loser, when am2+ owners will be able to upgrade to amd's latest and greatest cheaply and quickly, while intel will make you pay through your nose.

Seems like Phenom II will be a great upgrade to loads, a cheap alternative to some, and a great toy to OC with to few. This is an improvement.

[roofsniper]

Constructive criticism and frustration is 2 different things.

Alot of people seems to have a phailed dream here. Or yet another case of the hypes.

a lot of people also seem to think one review before the product is even launched means everything as well. we need more reviews and it needs to actually come out. the people here in the amd section have had luck with the last phenoms im sure they can do it again. i just don't understand how everyone can come to conclusions so quickly just because they see one thing. and as we all can see phenom II hits the marks it was intended to hit.

[BenchZowner]

lol Thats the new AMD fan cry, Benchmarks stopped mattering when AMD stopped winning and now the only important thing is gpu limited gaming benches, You could not make it up. Apparently The only important thing for a cpu now is running games on single gpu systems.

Unfortunately it's very very true for some people.
In the past they were shouting "Wow, AMD owns this this that and there".
Once AMD loses the battle "these benchmarks aren't real, they're worthless, etc".
Get your act together guys, which side are you on...
Red is red, blue is blue, and green is green.

I agree that some benchmarks are worthless since they are just synthetic and cannot reflect the real-world performance of the hardware, but there are also real-world benchmarks used ( and not used in this review ) that do reflect the real-life performance of the given hardware.
The fact that you're acting like there's not a single real-life benchmark in this review makes you seem arrogant at least.

{ obviously not pointed at you gallag, just found your post useful and a reason to express my thoughts when reading some posts }

I'm curious about who can OC higher and who uses less power. But I'd easily pick a Phenom II over a Q6600 based on these results.

It has been said that the average OC of a Phenom II 920/940 on air shall be around 3.6GHz to 3.8GHz, and in some "rare" cases 3.9 to 4GHz.
In the review you can see that the Phenom II doesn't perform way faster than the 400MHz disadvantaged Kentsfield ( Q6600 ).
The average OC of the Q6600 G0's is around 3.7GHz, and that's good enough to get equal or even beat in various applications a 3.9GHz or even a 4GHz Phenom II.
The same applies to the Q9450's and Q9550's... they can be easily overclocked to 3.7GHz+, and unless the Phenom's can do 4.1GHz to 4.2GHz on air, they don't have a chance.

I'm not saying that the Phenom II isn't a good processor, it's just not good enough to make people sell their LGA775 or LGA1366 rigs to buy a Phenom II based system.
It surely is a worthy upgrade for a AM2/AM2+ Phenom I owner.

So a site that nobody knows, an not even a "let's wait for credible sources", from the usual suspects??

First of all, every one has to start from something at a point.
Unless you think that AnandTech for example started off with a blast and got widespread in a day.

There are also some well-known and considered to be credible sites that give their readers a sh1tload of bullsh1t ( on purpose or not ).

on this forum, made by respectable members, that are a "little" different from the results on this review

You're registered here from 2006, but I don't really think that you're familiar with this forum.
First of all, you see that blue color in GoriLLakoS's username ?
That's because GoriLLakoS is in the X.I.P. ( a variation of the classic V.I.P., just Xtreme instead of very ).
The X.I.P. group is populated by credible people who work or have connections in various manufacturers, etc.
The fact that various well-known and credible people can and will easily vouch for GoriLLakoS should make you think a bit about what you said.
Nobody's forcing you to believe or accept these numbers from the review, but try to keep your eyes and mind open all the time, what you might consider fake or BS or stupid can easily be the truth.
Accusing somebody of something is one thing. Not a good thing, and not something that makes you look smart, bad ass or powerful, and of course doesn't make you a "man".
Recognizing that you were wrong and asking sorry is what separates the men from the boys.
Are you going to apologize to the authors if new reviews come out from other sites that you consider trustworthy and credible that agree with HWbox.gr's review numbers ?

since when 1680*1050 with max details are cpu limited or sth?

It can be CPU limited man.
Most of today's games are CPU limited even at 1920x1080 when you're not using any AA and/or AF.

If by "benches",some of you consider running superPi,wprime,pifast,everest built-in "tests","sisoftsandra" and similar ,then yeah these are about as useful as a chocolate teapot.

They are worthless for the average joe, but are valuable for some people called Benchmarkers.
If you're not a benchmarker, you can ignore them and skip the page.

Give me real world a/v compression

He just did.

give me archiving of files

He did, WinRAR compression test is there.
And it is in-line with real-life compression performance.

give me power draw at the wall socket in idle and full load,give me temperatures in idle and load

That's just wrong.
That method isn't the best of ideas, because the motherboard, RAM, and other components of each system draw different amounts of power, their efficiency is different, and of course even if you could ensure that both platforms components draw the exact same amount of power, and use the same PSU, you'd have to deal with the efficiency of the PSU, since each platforms power draw will be different and the PSUs efficiency varies depending on the power draw of the system.

Can you run this another encoding bench (HD Windows Media)?

Windows Media Encoder 9 is worthless.
Poorly coded resulting in bad efficiency, and the main problem is that it doesn't use any more than 2 cores.

Perhaps their newer product works better, Expression Studio 2, but I haven't thoroughly checked it out yet to see if it's utilizing more than 2 cores and if its efficient.

come on they are quick tests

They are "relatively" quick tests, but you have to run them at least 3 times to get a good image of the results and calculate the average numbers.

And it's harder to do, when you have more products to review.
And AN7 Overclocker has a lot of reviews to do

By "the good ones" can also be the ones that agree or fit your perception of what you want the numbers to be. Anand and Tech Report will probably have a huge amount of data but just like when the original Phenom was launched alot of people still didn't like the reviews because it painted a bad picture that certain people did not want to accept. Like it or not this review while being limited does provide a data point to where we should expect Phenom II to perform on "real" benchmarks. Good review guys.

For some people, your statement unfortunately stands true

Unfortunately you have not realized that you have insulted the amd cheerleading squad.

ROFL.
Funny but not beneficial at all.
It could easily lead to a flame war, so please don't do that again mate.

But we all know/knew that's never going to happen, Intel has sent amd back to the stone age and for the 3 years they had the success they had? It wasn't because amd is an innovative company, but due to the fact that Intel banked in an architecture which paid itself off by 100x at the very least.

Actually AMD innovated ( all alone and with the help of others, like Alpha, IBM, etc ) in the past, and can do that again.
They just need to get their asses down and work, work, work and work.
And maybe invest into some stuff and new personnel in certain areas.

And there's always a chance of AMD being ahead ( in the lead ) in the future.
And guess what. It's not a bad thing, on the opposite, it's a good thing.

So, my advice? Next time, lie a little (in favor of another mediocre device) and you'll see how quickly you'll be praised by the amd cheerleading squad. I'd do it.

Delete this please.

i don't recall anyone saying that phenom II will outperform i7. i do remember people saying tho that i7 isn't 5x better than phenom II as the price acts like it is.

As a new product and since the market is all about competition ( good or bad ), and you need to grab the performance crown every once in a while if you want to stay in the market.

By the way, the Core i7 920 doesn't cost 5x the phenom II's price, unless you live in another world. Do you ?

phenom 2 can do better i have seen it with my own eyes. and i know that it has the function of being overclocked as well which is why im wondering how it can be called a full review. its funny all the things that phenom II should be better at haven't been posted in this review. pricing, how well it overclocks, and power consumption. instead all we are seeing is benchmarks that don't even matter and how a cpu performs on that.

Let's just assume that the average overclock of the Phenom II 920 and 940 will be 4.1GHz on air.
The average OC of the Q6600's G0 is at 3.7GHz.
In this case, the Phenom II will beat the Q6600 in 70% of the applications.
The average OC of the Q9450's is around 3.8GHz, and that's enough to catch up with and beat a Phenom II @ 4.1GHz in various applications.

I think you saw reviews of another planet... if you even had the slightest idea how boring it is to watch all them gamedemos, benches,... To rerun them 3-5 times to obtain an average result... if the numbers don't please ya plz don't post anymore as it's getting very annoying what you and some other are trying to force here...

I find it a very objective review and it's one of the first too. It seems the guys are still working hard on the review and are adding results all the time. so it's not done yet !!!

Maybe it doesn't include all the applications you run. But feel free to link me to a website that does all your stuff in their reviews... people want to see what they are known to see... 3dmarks, Superpi's, games,... boring for you ? I think your stuff is even more boring even though they are your real world apps...

The review shows clearly the weaknesses and the strengths to me of the New Phenom II@stock... Is PII overhyped : yep it is to me. AMD's for sure is on the way back CPU wise, but not in a brutal way as some would have hoped (like they did with the GPU's...)

PhenomII is a decent CPU, is it a revolution ? Nope it's a nice evolution...but nothing earthshattering...If you don't see that too bad... it's a good upgrade for most AM2 users, can it compete with I7 ? ( as this seem to be the CPU of interest for most flamers here)... hardly think so... but it will extend the life cycle of AM2 socket for current AMD users which is a very good thing.

And don't repeat a zillion times how expensive I7 is... the i920 retails a tiny bit higher as your beloved PII 940 here... just the mobo and ram are more expensive ( this is if you like triple channel kits) What will you do with AM3 ? As you will have to use the same DDR3 modules as we are using now on socket 775 or I7 platform...I can buy a decent 4GB PC12800C9 kit for 120 euro's, I7 mobo's are retailing around 220 euro's...

Only thing that keeps me from moving to I7 is the heatoutput when doing 4Ghz ops...I'll await Bulldozer or the I7 die shrink...PII it will not be... unless to mess around with it

Give the HWbox guys some time to complete the review and add some more stuff instead of being so rude and ungratefull...

I really hope the AMD fanboys will get some "real" numbers from other sites, that are being referred to as more "reliable" or let's just say that they publish results that PLEASE you...

Good job Team Hwbox !! Plz add some more games and then finally the OC'ing and thermal/consumption part...

Nuff said...

Agreed on most if not everything

The best gaming CPU is the one that can handle the load when the game is working hard. All games differs in how much CPU power they need depending on what happens in the game. If the game is in a actions scene then physics, fights etc makes the processor work much harder, process more memory and there is where you need capacity from the CPU.
By using a slow video card or higher resolution it is possible to filter away fps numbers in the higher areas (the video card will cut the high fps values) because those aren't important for game play anyway (the game will slow down on the lower areas). The older Phenom was good if you did this filtering but it was rather hard to filter values in right areas. If the video card is to slow that will bottleneck the game all the time and then you don't see the lower fps areas.
Phenom II has make it easier to see the lower areas where it is strong, the margin is wider because it is good (better compared to Phenom) on areas where there is little action, maybe only one core need's to work and most data needed can be handled by the big L2 cache on Core 2 processors.
Making the filter more narrow (higher resolution and/or slower video card) will increase the lead from Phenom II if the test shows this behavior, of course it depends on how the game works. Single threaded games that doesn't use much data will probably get best performance for a high clocked C2D.

Testing average fps has the effect of counting frames from different areas in the game

Man, give us a break with your crap.
Sorry XS staff, I couldn't resist.

[PetNorth]

I've made a summary with benches run till now (jumping crappy Everest and Sandra):

[STEvil]

Thread locked.

edit - re-opened by request.

[Leeghoofd]

Let's keep the thread clean guys and focus on the extra benches to be added soon by the HWboxcrew...

Go go Greece !!

Thx blond admin with the nice curves

[OverClocker_gr]

Thanx STEvil


ok. Lets sum up.

First of all later today we will add
UT3
GTA IV
ETQW

we decided that clock to clock comparison will be @ 3700mhz.

Besides that, a Phenom X4 9950BE clocked at 3ghz will be compared to deneb at same speed to see if the only thing that amd did is die shrink

[duploxxx]

Thanx STEvil


ok. Lets sum up.

First of all later today we will add
UT3
GTA IV
ETQW

we decided that clock to clock comparison will be @ 3700mhz.

Besides that, a Phenom X4 9950BE clocked at 3ghz will be compared to deneb at same speed to see if the only thing that amd did is die shrink

while oc compare pls give also the nb a speedbump to 2,2ghz that will be the launch speed of the am3 proc and easy oc gain, no idea why this was lowered in this 920-940 original release.

[OverClocker_gr]

nb and ht link are @ 2250

[sundancerx]

agena vs deneb @ 3.0

deneb vs kentsfield vs yorkfield vs bloomfield @ 3.7

pls dont forget temps (if possible even without temps for deneb) system wattage, and more game s review pls.

edit: dont forget to mention if o;clock doen on stock cooler, or if after-market cooler is used, and how much. i assume 3.7 will be done on air. thnks

[sundancerx]

Thanx STEvil


ok. Lets sum up.

First of all later today we will add
UT3
GTA IV
ETQW

we decided that clock to clock comparison will be @ 3700mhz.

Besides that, a Phenom X4 9950BE clocked at 3ghz will be compared to deneb at same speed to see if the only thing that amd did is die shrink

great thanks.

[sundancerx]

I've made a summary with benches run till now (jumping crappy Everest and Sandra):

great job man. you did not include i7?

[Chad Boga]

This review was interesting and hints at the Phenom II's performance levels, but I am anxiously awaiting to see if reviews from The Tech Report and Anandtech make things any clearer or not.

I also thought the i7 results were bizarre and whilst I want to see what happens when The Tech Report and Anandtech compare i7 to PhII, it does support my theory that the i7 is an unbalanced CPU at this stage, reminiscent of the first generation Pentiums for mine.

[gOJDO]

Well done guys! Thank you for the comprehensive review. It gives us a more clear image about the potential of PII. Looking at the numbers I'm glad I wasn't waiting for PII and upgraded my Q9450 to a i7 920.

Just keep up the good work and don't bother with the AMD fanboy trolls. I'll ask you, since you are going to make a clock-to-clock comparisons, to include some single-threaded apps or games like FEAR and Super Pi. That's the best way to see the IPC advantage of Deneb over Agena.

BTW, I saw PII in Plaisio.gr's prices list. Is it available for buying?

[OverClocker_gr]

thanx man

i figure out something



i think that all we agree that SiSoft Sandra is useless.
Right?

[s3rv3r]

Try to use with 1920x1080 in games and if is possible with SLI of GTX 260-216.

[OverClocker_gr]

Try to use with 1920x1080 in games

cant,monitor limitations

and if is possible with SLI of GTX 260-216.

@ yorkfield and kentsfield its imposible due to mobo.

260-216 at the comparison, runs @ 710/1100/1446

[Chad Boga]

i think that all we agree that SiSoft Sandra is useless.
Right?

I do think it is a worthless benchmark, but many people like it because they have it at home so can do a direct comparison.

But if a Sandra result shows that i7 or PhII is 5 times faster than your computer at home, on just what applications would it be 5 times faster in real usage?